The Social Science Journal 49 (2012) 150–154

Contents lists available at SciVerse ScienceDirect

The Social Science Journal

j our na l ho me p age: www.elsev ier .com/ locate / sosc i j

Ethnic concerns and Latino party identification

Diana Evansa, Ana Francob, J.L. Polinardb, James P. Wenzelb, Robert D. Wrinkleb,∗

a Trinity College, Hartford, CTUSAb University of Texas-Pan American, USA

a r t i c l e i n f o

Article history:Received 19 May 2010Received in revised form 19 August 2011Accepted 27 August 2011Available online 18 April 2012

Keywords:

a b s t r a c t

The accelerated growth of the Latino population in the United States has made Latinos acoveted addition to each major political party’s base. In this paper we examine the influenceof ethnic concerns on the party identification of Latinos in the U.S. In contrast to previousstudies, we account for Latinos’ perceptions of the political parties’ concern for their eth-nic interests, allowing such interests to be self-defined. In a multinomial logit analysis ofpooled data from three surveys of Latinos taken in 1999, 2004, and 2006, we find such per-ceptions do affect Latino partisanship, along with variables such as nativity and country of

LatinoParty identificationEthnic concerns

origin or ancestry. We also find a tendency toward independence among Latinos. Finally,we find movement toward the Democratic Party in 2004, once ethnic concerns are takeninto account. One implication of the findings is that the party that can best persuade Latinosof their concern for their interests is the party most likely to gain their loyalties; indeed,the parties must earn those loyalties.

ocial S

© 2011 Western S

1. Introduction

Most Latinos (sans Cuban-Americans) have long beenfound to be firmly Democratic in their partisan identifica-tion (Alvarez & Garcia Bedolla, 2003; de la Garza, DeSipio,Garcia, & Falcon, 1996; Garcia, Hero, Pachon, & Garcia,2000; Grebler, Moore, & Guzman, 1970), despite recentefforts by Republican strategists to convert Latinos to theircause (Basler, 2008; de la Garza & Cortina, 2007; Gimpel,2004). Nevertheless, Hajnal and Lee (2011) find that thepercentage of Latinos who identify as independents ordecline to state any identification exceeds the percent-age identifying as Democrats and Republicans put together(p. 7).

In this paper, we investigate the determinants of Latinoparty identification as measured by surveys from 1999,

2004, and 2006. We hypothesize that Latinos’ partisanidentification over this period is affected by ethnic groupconcerns broadly defined as perceptions of which party has

∗ Corresponding author. Tel.: +1 956 665 3679.E-mail address: rdwe116@panam.edu (R.D. Wrinkle).

0362-3319/$ – see front matter © 2011 Western Social Science Association. Publidoi:10.1016/j.soscij.2011.08.013

cience Association. Published by Elsevier Inc. All rights reserved.

more concern for Latinos. There is evidence that having asense of an ethnically linked fate affects Latinos’ partisanidentification in favor of the Democratic Party (Hajnal &Lee, 2011), as does the experience of being the target ofdiscrimination (Cain, Kiewiet, & Uhlaner, 1991). Given thisevidence on selected issues and experiences, we test thehypothesis that Latino partisanship is broadly affected byperceptions that a party is concerned with Latinos. More-over, the Pew Hispanic Center found considerable partisanmovement among Latinos between 1999 and 2006 (Lopez& Minushkin, 2008). We examine that movement over timeand locate the approximate date and direction of partisanchange.

2. The literature

The literature indicates that Latino partisanship is influ-enced by a combination of ethnic variables and factors that

also influence Anglo partisanship. Among the latter, incomeand education do not always have the same effects for Lati-nos as for Anglos. Some have found income or personalfinances to be unrelated to party identification (Alvarez &

shed by Elsevier Inc. All rights reserved.

cience J

GBmR2HbGtBi

nD(2eaUmiB&

bdDgadm

DddRi&tt2tGsmem

wtcqBicat

ttw

ability to carry on a conversation in English; and pri-mary language spoken at home. Scale scores range from3 to 13; higher scores indicate greater acculturation.2 The

1 The exact coding is not important for the multinomial logit analysis

D. Evans et al. / The Social S

arcia Bedolla, 2003; de la Garza & Cortina, 2007; Dutwin,rodie, Herrmann, & Levin, 2005); others have found theore typical positive relationship between income and

epublican identification (Bowler, Nicholson, & Segura,006; Cain et al., 1991; Fraga, Valenzuela, & Harlen, 2009;ajnal & Lee, 2011; Kelly & Kelly, 2005). Education haseen found variously to be unrelated to partisanship (de laarza & Cortina, 2007; Fraga et al., 2009); weakly yet posi-

ively related to Republican identification (Alvarez & Garciaedolla, 2003); and positively related to Democratic Party

dentification (Hajnal & Lee, 2011).Ethnic variables loom large in Latino partisanship. Lati-

os with Mexican and Puerto Rican roots tend to beemocrats, but Cuban Americans tend to be Republicans

Alvarez & Garcia Bedolla, 2003; de la Garza & Cortina,007; Dutwin et al., 2005; Fraga et al., 2009; Greblert al., 1970). Latino partisanship is also affected by vari-bles related to acculturation, such as nativity, years in the.S., and age, with non-Cuban Latinos and the native-bornore likely to identify as Democrats as they are increas-

ngly exposed to the U.S. political system (Alvarez & Garciaedolla, 2003; Cain et al., 1991; Dutwin et al., 2005; Hajnal

Lee, 2011).Additionally, Wong (2000) finds English proficiency to

e related to the acquisition of partisanship, although sheoes not examine the direction of the effect. Similarly,utwin et al. (2005) argue that language is the best sin-le measure of acculturation. However, they decline todvance a hypothesis regarding its effect on partisanshipue to conflicting indications in the literature; in theirodel, they find that it has no impact at all.There is evidence that non-Cubans’ preferences for the

emocratic Party stem in part from being the target ofiscrimination (Cain et al., 1991; Hajnal & Lee, 2011)ue to the Democrats’ pro-racial equality policies and theepublicans’ frequent anti-immigration rhetoric and pol-

cy proposals (Bowler et al., 2006; Hero, Garcia, Garcia, Pachon, 2000). In addition, some scholarship examines

he impact of specific ethnically related issues, includinghe sense of linked fate with other Latinos (Fraga et al.,009; Hajnal & Lee, 2011); attitudes toward immigra-ion (Fraga et al., 2009) and affirmative action (Alvarez &arcia Bedolla, 2003), all of which are linked to partisan-hip. Moreover, those who identify ethnically as Latinos areore likely to be Democrats (Dutwin et al., 2005); however,

vidence on partisan tendencies of naturalized citizens isixed (Dutwin et al., 2005; Fraga et al., 2009).While the issues of concern to Latinos vary over time, as

e show below, several recent surveys make it possibleo measure the impact over time of Latino-centric con-erns on partisan identification; those surveys include auestion asking which party has more concern for Latinos.y defining Latinos’ interests generally (in effect, allow-

ng the respondent to apply his or her own definition), wean comprehensively determine whether ethnic concernsffect Latinos’ partisanship by examining their impact overime, even as the specific issues shift.

The impact of such concerns on African American par-isanship has been examined; Hajnal and Lee (2011) findhat blacks’ partisanship is affected by perceptions ofhich party works harder on issues that blacks care about.

ournal 49 (2012) 150–154 151

However, a comprehensive, subjectively defined measurehas not been used to estimate Latino partisanship. Thisresearch note fills that gap.

3. Hypotheses and model

Our major hypothesis is that Latinos are more likelyto identify with the party that they deem to have themost concern for them. Drawing on the literature citedabove, we also expect that native born Latinos and Lati-nos of non-Cuban origin are more likely to be Democratsthan independents or Republicans. However, we have nospecific expectations for income and education, given themixed findings of previous research. In addition, givenWong’s (2000) finding that language affects the acquisitionof partisanship, we include it in the model. But followingDutwin et al. (2005), we advance no specific hypothesisregarding its effect. We also expect that women tend toidentify as Democrats, reflecting the gender gap in thelarger society.

Finally, although Latinos have identified strongly asDemocrats over time (de la Garza & Cortina, 2007), therewas change in Latino partisanship between 1999 and 2006,when the Pew Hispanic Center found that Latinos movedtoward the Republican Party but then reversed course,moving more strongly than before into the ranks of theDemocrats (Lopez & Minushkin, 2008). To test for evidenceof this change, we include dummy variables for the yearsof the surveys that supply the data for this analysis.

4. Data and methods

The data are drawn from surveys of Latinos taken in1999, 2004, and 2006 (Appendix A). We create a pooleddataset, using questions common to each survey; the word-ing of these questions is identical. While pooling the datareduces the number of variables that may be used, it allowsus to model changes in partisanship over time independentof other variables.

The dependent variable consists of responses to thisquestion: “In politics today do you consider yourself aRepublican, a Democrat, an Independent, or member ofsome other party?” in the analysis, Republican = 1; Demo-crat = 2 and independent = 3.1

For nativity, native-born Latinos are coded 1, otherwise0. National origin is measured as a set of dummy vari-ables: Mexican, Puerto Rican, Cuban, and “other” Latino.The measure of acculturation is modeled on Marin, Sabogal,Marin, Otero-Sabogal, and Perez-Stable (1987); it con-sists of an additive scale of three measures of English useincluding: ability to read a book or newspaper in English;

used here, as the estimates for each value of the dependent variable arereported in relationship to one other specific value.

2 This scale produced a Cronbach’s alpha of .811, indicating a highdegree of reliability. The mean score on this scale is 7, indicating a mod-erate degree of acculturation.

cience Journal 49 (2012) 150–154

U.S. are more likely to have experienced discrimination.We speculate that such experiences are also more com-mon among the least acculturated—by this measure, thosewho primarily use Spanish—even if they are native-born.

152 D. Evans et al. / The Social S

coding of income, education, and gender are described inAppendix A.

The key independent variable is measured by the fol-lowing question: “Which party has more concern forLatinos” – Democrats, Republicans, or no difference? Weconstructed three dummy variables from these responses.“Republicans more concerned” was coded 1, otherwise 0,and so on through the three responses. A possible concernwith this variable is that those who are already Democratssimply assert that the Democratic Party is more concernedwith Latinos, thus rationalizing their choice of party; ifthat were the case, the party concern variable would beendogenous. To assess this possibility we modeled “partymore concerned” in each survey year, using as indepen-dent variables all of the specific ethnically related issuesin that year’s survey. Those issues change from survey tosurvey. To the extent that the respondents’ positions onthose issues explain their choice of the party they considermore concerned, we can have confidence in that variableas a measure of ethnic concerns.

In the interests of space we briefly summarize theresults, which are available from the authors.3 In 1999,the independent variables that are positively related to thedependent variable (indicating a tendency to choose theDemocratic Party as most concerned) are personal expe-rience with discrimination, the sense that Latinos wouldbenefit from working together, and support for diplomaticrelations with Cuba. In the 2004 survey, the significant vari-ables are opposition to George Bush’s immigration proposaland the sense that discrimination is a major problem. In2006, they are the sense that the immigration marches of2006 mark the beginning of a new Latino movement, thebelief that the Democratic Party has the best position onimmigration, and the sense that discrimination is a majorproblem. These results are consistent with the Demo-cratic Party’s long-standing pro-civil-rights stance (withthe exception of Bush’s immigration proposals). Therefore,it is likely that attitudes on these issues are not mere ratio-nalizations of Democratic Party identification but, in fact,help to drive such identification.4

5. Results

Fig. 1 displays a plot of the changes in Latinos’ parti-san identification over the three surveys. Overall, while notstatic, Latino partisanship does not change dramatically.Yet there are noticeable differences in the marginal distri-butions: In 1999, 46.02% identified as Democrats, 20.46%as Republicans and 33.53% as independents. In 2004,Republicans made some gains, with 26.57% identifying as

Republicans. This increase came in almost equal percent-ages (about 3% each) from Democrats and Independents,with Democratic identifiers at 43.24% and Independents

3 We report results for an ordered logit analysis, where the cod-ing for the dependent variable is 1 = Republican, 2 = no difference, and3 = Democrat. The results for a multinomial logit analysis of the samemodel are virtually identical.

4 Moreover, the correlation between party identification and percep-tions of party concern (using Kendall’s tau) is .39; the two variables areindependent beyond the .001 level of confidence.

Fig. 1. Latino party identification 1999–2006.

at 30.19%. However, in 2006, Republicans lost about 7%from their 2004 total with those numbers reverting notto Democrats, but to Independents, at 37.69% of Latinos;their Democratic identification remained relatively staticat 43.16%. Thus, Republican gains were temporary, whileLatino independence grew.

Latinos’ responses to the “party more concerned” ques-tion are as follows – Democrats: 42.8%; Republican: 12.9%;no difference: 44.17%. Fig. 2 displays the bivariate rela-tionship between Latinos’ party identification and theirperceptions of the party more concerned with Latinos.Those who think one party or the other is more concernedare more likely to identify with that party. Nevertheless,at least 20% of such respondents identify as independents,and more than half who see no difference still identify witha party.

To determine the role played by explicitly ethnic con-cerns relative to the other variables specified above, weestimated a multinomial logit model of Latinos’ partisanidentification. The results of the analysis are shown inTable 1. In the first two columns, the base category is iden-tification with the Republican Party.

In the first column, we see that several variables aresignificant. If one believes that the Republican Party hasmore concern for Latinos, one is more likely to identify asa Republican; the reverse is true if one believes that theDemocratic Party has more concern; the omitted categoryis “no difference.” Latinos who are more acculturated areless likely to identify as Democrats than as Republicans.However, native-born respondents are more likely to iden-tify as Democrats. The latter result may reflect Cain et al.’s(1991) finding that Latinos who have lived longer in the

Democratic Republican Independent

0

10

20

30

40

50

60

70

80Democratic Party

Republican Party

No Difference

Party Identification

Pe

rce

nta

ge

(%

)

Fig. 2. Percentage partisan identification among Latinos by party moreconcerned, 1999–2006.

D. Evans et al. / The Social Science Journal 49 (2012) 150–154 153

Table 1Determinants of Hispanic party identification.

Independent variable Coefficienta (Std. Error)

Democrats v. Republicans Independents v. Republicans Independents v. Democrats

Republicans more concerned −2.05*** (.49) −1.23*** (.31) .81 (.51)Democrats more concerned 2.66*** (.32) .94*** (.33) −1.72*** (.23)Acculturation −.12* (.06) −.06 (.05) .06 (.05)Income −.09 (.08) −.02 (.07) .07 (.06)Gender .01 (.27) .19 (.25) .18 (.22)Education −.04 (.08) −.05 (.08) −.01 (.07)Native born .70* (.38) −.04 (.36) −.73** (.34)Mexican .55* (.30) 1.02** (.35) .47 (.31)Puerto Rican .32 (.50) −.05 (.56) −.37 (.49)Other ethnic .75** (.36) .75* (.40) −.01 (.37)2004 dummy .73*** (.23) .18 (.23) −.54*** (.21)2006 dummy .10 (.26) −.11 (.25) −.21 (.17)Constant .07 (.51) .31 (.49) .24 (.43)

Number of observations 4820Wald chi-square 190.34*** (24 d.f.)Pseudo R2 0.18

a Coefficients are multinomial logit estimates.* Significant at .10.

**

CEmoa

dmtn“I

DhabaAbi

weoa

6

bi

i

Significant at .05.*** Significant at .01.

onversely, those who are foreign-born but use mainlynglish might experience less discrimination and thus beore likely to be Republicans. Finally, if one is of Mexican

r “other” national origin, one is more likely to identify as Democrat than Cuban-Americans and Puerto Ricans.

The second column shows the results of the Indepen-ent v. Republican analysis. The results for the “partyost concerned” variable are similar to—albeit weaker

han—those in the first column. Acculturation has no sig-ificant effect, nor does nativity. If one is of Mexican orother” national origin, one is more likely to identify as anndependent than as a Republican.

The third column compares independents andemocrats. Those who believe the Republican Partyas more concern for Latinos are no more likely to identifys independents than Democrats. However, those whoelieve the Democratic Party has more concern for Latinosre significantly more likely to identify with that party.cculturation has no significant impact, but again, native-orn Latinos are more likely to identify as Democrats than

ndependents.Latinos overall were significantly more likely to identify

ith the Democrats in 2004 than in 1999; there was noffect for 2006 relative to 1999. The demographic variablesften found to be related to partisanship are not significantfter ethnic issues are taken into account.5

. Discussion and conclusions

While numerous scholars have produced a valuableody of knowledge concerning the dynamics of Latino party

dentification, we have demonstrated that there is consid-

5 Income and education are not highly correlated: Kendall’s tau = .36; its not significant.

erable value in including Latino-centric concerns in suchmodels. In virtually all of our estimates, perceptions ofwhich party has more concern for Latinos significantlyinfluence Latinos’ party identification. That variable fails toreach significance only for “Republicans more concerned”in the independents v. Democrats model, an unremarkableresult.

With respect to country of origin, we find thewell-known differences between Latinos of Cuban andnon-Cuban origin, with Mexicans and “others” particularlyunlikely to identify as Republicans. However, the failure ofthose variables to attain significance in the independents v.Democrats model suggests that among non-Cubans, thereis no significant preference for Democrats over indepen-dence once ethnic concerns are accounted for. In the othercomparisons, the ethnic variables have an independenteffect; yet in two of these cases, the coefficients are signif-icant only at the .1 level of confidence. This suggests thatthe Democratic Party does not have a lock on Latinos’ par-tisan loyalties even among those ethnic groups normallyfound to be most strongly inclined to identify with it. More-over, acculturation only benefits the Republican Party ascompared with Democrats; more acculturated Latinos areas likely to be independents as Republicans or Democrats,again indicating a tendency toward independence amongLatinos (Hajnal & Lee, 2011). All else equal, however, thenative-born are strongly inclined toward the DemocraticParty. Finally, the trend over time toward the DemocraticParty, which manifests itself in 2004, is strong. However,the contrast with 1999, the omitted category, does notappear in 2006, a surprising result given the national tidetoward the Democrats in that year. The Republican surge

among Latinos in the early 2000s may have been a functionof George W. Bush’s immigration proposals, a variable thatcontributes to the choice of the Republican Party as moreconcerned in the individual survey year analyses described

cience J

154 D. Evans et al. / The Social S

above. Once that is taken into account, the year variablereflects a trend toward Democrats.

The increase in independence among Latinos is notreflected in the larger population of the U.S. during thesame time period. In national surveys, the Pew ResearchCenter found the percentage of the population not affil-iated with a party unchanged at 30% between 2004 and2006 (Pew Research Center for the People and the Press,2009). For Latinos in 2006, the level of independence is37%, nearly 25% higher than for the general population. Thetrend toward independence is one that both parties willhave to deal with, one that might present opportunities foreither party to make gains in individual elections, depend-ing on short term forces. At the very least, the importanceof Latino-centric issues to partisan identity suggests that ifthe major political parties want to win the firm allegianceof Latinos, they must earn it.

Appendix A.

The 1999 data were collected in a nationally repre-sentative sample survey conducted by telephone by theWashington Post, the Henry J, Kaiser Family Founda-tion and Harvard University. The survey had 2,417 Latinoadults 18 years and older and was weighted to provide anationally representative sample. This survey also includednon-Latinos, but only Latinos are analyzed in this paper.Respondents were coded as Latino if they answered yes tothe question, “Are you, yourself of Hispanic or Latin ori-gin or descent, such as Mexican, Puerto Rican, Cuban, orsome other Latin background?” The respondents included312 Cubans, 318 Puerto Ricans and 818 Mexicans, withthe remaining 593 coming from Central or South Americanbackground.

The 2004 data come from the joint Pew Hispanic Cen-ter/Henry J. Kaiser Family Foundation study: “NationalSurvey of Latinos: Politics and Civic Participation.” Thisstudy included 2,288 respondents and was weighted toprovide a nationally representative sample of Latinos. Thesample included 1,110 Latinos of Mexican descent, 282 ofPuerto Rican descent, 421 Cubans and 447 of other nationalorigin background.

The 2006 National Survey of Latinos: The ImmigrationDebate was conducted by telephone by the Pew HispanicCenter among 2,000 Hispanic adults, 18 years and older,who had the option to respond in Spanish, English, or acombination of the two languages (Pew Hispanic Center,2006)6. The sample was weighted to provide a nation-

ally representative sample. The respondents included 713respondents of Mexican origin, 300 of Puerto Rican Origin,238 of Cuban and the remainder from other Latin Americanor European background.

6 The Pew Hispanic Center and (for the 2004 survey) the Kaiser Fam-ily Foundation bear no responsibility for the interpretations offered orconclusions made based on analysis of the 2004 and 2006 surveys usedhere.

ournal 49 (2012) 150–154

The control variables in the multivariate model arecoded as follows: for gender, female = 1; male = 0; foreducation, 1 = none through grade 8; 2 = high school incom-plete; 3 = high school grad or GED; 4 = business, technicalor vocational school; 5 = some college, no 4-year degree;6 = college graduate; 7 = post-graduate training or profes-sional. 9 = “refused,” and is treated as missing. Income = 1 ifless than $50,000 and 2 if more than $50,000.

References

Alvarez, R. M., & Garcia Bedolla, L. (2003). The foundations of Latino voterpartisanship: Evidence from the 2000 election. Journal of Politics, 65,31–49.

Basler, C. (2008). White dreams and Red votes: Mexican Americans andthe lure of inclusion in the republican party. Ethnic and Racial Studies,31, 123–166.

Bowler, S., Nicholson, S. P., & Segura, G. M. (2006). Earthquakes and after-shocks: Race direct democracy, and partisan change. American Journalof Political Science, 50, 146–159.

Cain, B. E., Roderick Kiewiet, D., & Uhlaner, C. J. (1991). The acquisitionof partisanship by Latinos and Asian Americans. American Journal ofPolitical Science, 35, 390–422.

de la Garza, R. O., DeSipio, L., Chris Garcia, F., & Falcon, A. (1996). Willthe real Americans please stand up? Anglo and Mexican-Americansupport of core American political values. American Journal of PoliticalScience, 40, 335–351.

de la Garza, R. O., & Cortina, J. (2007). Are Latinos republican but just don’tknow it? The Latino vote in the 2000 and 2004 presidential elections.American Politics Research, 35, 202–223.

Dutwin, D., Brodie, M., Herrmann, M., & Levin, R. (2005). Latinos andpolitical party affiliation. Hispanic Journal of Behavioral Sciences, 27,135–160.

Fraga, L. R., Valenzuela, A. A., & Harlen, D. (2009, March 19–21). Patternsof Latino partisanship: Foundations and the prospects for change. InDelivered at the annual meeting of the Western Political Science Associ-ation Vancouver, BC, Canada.

Garcia, F. C., Hero, R., Pachon, H. P., & Garcia, J. (2000). Latino participation,partisanship, and office holding. PS: Political Science and Politics, 33,529–534.

Gimpel, J. G. (2004). Losing ground or staying even: Republicans andthe politics of the Latino vote. Center for Immigration Studies,http://www.cis.org/articles/2004/back1004.html

Grebler, L., Moore, J. W., & Guzman, R. C. (1970). The Mexican Americanpeople: The nation’s second largest minority. New York: The Free Press.

Hajnal, Z., & Lee, T. (2011). Race, immigration and (non)partisanship inAmerica. Princeton: Princeton University Press.

Hero, R., Garcia, F. C., Garcia, J., & Pachon, H. (2000). Latino participation,partisanship, and office-holding. PS: Political Science and Politics, 33(3),529–534.

Kelly, N. J., & Kelly, J. M. (2005). Religion and Latino partisanship in theUnited States. Political Research Quarterly, 58, 87–95.

Lopez, M. H., & Minushkin, S. (2008, July). 2008 National survey of Latinos:Hispanic voter attitudes. Washington, DC: Pew Hispanic Center.

Marin, G., Sabogal, F., Marin, B. V., Otero-Sabogal, R., & Perez-Stable, E. J.(1987). The development of a short acculturation scale for Hispanics.Hispanic Journal of Behavioral Sciences, 9, 183–205.

Pew Hispanic Center (2006). The 2006 National Survey of Latinos: TheImmigration Debate. http://www.pewhispanic.org.

Pew Research Center for the People and the Press. (April 2009). GOP party

identification slips nationwide and in Pennsylvania. Washington, DC:Pew Hispanic Center.

Wong, J. S. (2000). The effects of age and political exposure on the develop-ment of party identification of Asian Americans and Latino immigrantsin the United States. Political Behavior, 22, 341–371.